Basic principles of photon detectors used in Astronomy...May 13, 2009 · Reinhold Dorn -Basic...

1st EIROforum School on Instrumentation

13 May, 2009 1Reinhold Dorn -Basic principles of photon detectors used in Astronomy

Basic principles of photon detectors used in Astronomy

Reinhold J. Dorn

ESO Instrumentation Division



these notes will focus on detectors used in Astronomy with awavelength coverage from the UV to the near infrared.

Detectors are 2-dimensional and detect photons or intensityso one cannot measure color directly.

For wavelength longer than 20 microns the low energy photonscannot be detected directly. Those detectors measure thephysical effects such as heat or a change in resistance.

We will talk about:

1. Optical detectors are usually CCDs and CMOS devicesbased on silicon (SI).

2. Infrared detectors are based on IR detector materialsuch as HgCdTe or InSb hybridized to a siliconmultiplexer.

There are other technologies as APD (photon counting), wavefrontsensors for Adaptive Optics and STJs (superconducting tunnelingjunctions, those can measure 3D).

There are many ways to sense light, but ..



There are many ways to sense light, but ..

not all of the light gets through the atmosphere to ground-based telescopes

Except for visible, some NIR and radio waves, all other EM radiation is blocked by the atmosphere

Blocking is caused by H2O –vapor, Ozone (O3), oxygen (O2) and Carbon dioxide (CO2)

Other observations must be made from space (i.e. Hubble, JWST, Satellites)



Sir Isaac Newton identified the problem 300

years ago

“For the Air through which we look upon the Stars, is in a perpetual Tremor”.... “But these Stars do not twinkle when viewed through Telescopes which have large apertures”... “The only Remedy is a most serene and quiet Air, such as may perhaps be found on the tops of the highest Mountains above the grosser Clouds.“ (Isaac Newton, 1730)



Detectors used in Astronomy are usually made out of semiconductor materials



The basic mechanism behind the CCD and IR detectors is the principle of the photoelectric effect.

Planck said that radiation from a heated sample is emitted in discrete energy levels, called “quanta”. The Energy is hv, where h is the Plank constant and v the frequency. Soon after Plank, Einstein interpreted an experiment which proved the discrete nature of light.

Lets assume that a UV photon of one wavelength hit the surface of a metal plate in vacuum.

The electrons in the metal absorb the energy of the photons and some receive enough to be ejected into the vacuum.

Metal

Slope=h

Kinetic energy

Electron

Frequency (v)

Em

Photon E=hv

By measuring the energy of the escaping electron a plot can be made of maximum kinetic energy as a function of frequency of the photons.Ekin of the electron is independent of the light intensity.

qhmvEkin 2

2

1

where Ekin is the maximum Energy of the ejected electron, q is the electron charge and Φ (volts) is the characteristics of the metal used. q Φ is the minimum required energy for the electron to escape from the specific metal (workfunction)



What are Semiconductors ?

Elemental semiconductors are column IV elements (e.g., Si, Ge) Outermost shell contains 4 electrons The four electrons form perfect covalent bonds with four

neighboring atoms creating a crystal lattice

Si - IV semiconductorHgCdTe - II-VI semiconductorInGaAs & InSb - III-V semiconductors

Silicon Crystal Structure

Electrons are trapped in the crystal lattice

– by electric field of protons

Light energy can free an electron from the grip of the protons, allowing the electron to roam about the crystal

– creates an “electron-hole” pair.

The photo charge can be collected and amplified, so that light is detected

The light energy required to free an electron depends on the material.



Absorption of photons in a semiconductor - Valence & Conduction Bands

In a crystal lattice, the allowed bands of electrons can be described by valence and conduction bands (this is similar to quantum orbits of electrons in Hydrogen).

valence band = "ground states" that are normally completely filled

conduction band = "excited states" that are normally completely unfilled, electron inthe conduction band can move if there is electric field

no electrons between valence and conduction bands

Insulator

Valence band

Conduction band

Semiconductor Metal

Eg

Eg

Eg=0

Eg(Insulator) >> Eg(Semiconductor)>> Eg(Metal)

Eg is the bandgapenergy between the valence and the conduction band.



cutoff

KT

200max

)(

24.1)(

eVbandgapbandgap

cutoffEE

hcum

How do we move electrons from valence band to conduction band in semiconductors?

There are two methods to move electrons from the valance band to the conduction band:

This is the origin of dark current and why we have to cool detectors

Photon energy (hν) > band gap energy (Eg) => photo-electron can jump into conduction band

This is basically why semiconductors are used for astronomical observations.

The longest wavelength a detector is sensitive is the cutoff wavelength λcutoff.

Photoelectric effect by photons absorbed by the semiconductor

kT

ENn

g

e2

exp

ne → Number of electrons promoted

across the gap (= no. of holes in the valence band)

N → Number of electrons available at the top of the valance bandfor excitation

By thermal excitation of electrons in the valance band (intrinsic)


13 May, 200910Reinhold Dorn -Basic principles of photon detectors used in Astronomy

Material Symbol Ebandgap λcutoff

Silicon SI 1.12 1.1

HgCdTe HgCdTe 1.0-0.09 1.24-14

Indium Antimonide InSb 0.23 5.5

Arsenic doped

Silicon

Si:As 0.05 25

In a semiconductor ions of another material are being by implanted. The implant makes a collection site for the photo-charge. This forms a p-n junction or diode which is biased to produce an electric field. An electron-hole pair is separated by the E-field and the electrons are accumulated

on the diode. Then you can measure the voltage across the diode which is proportional to the number

of electrons.

Conduction band

Valence band

Electron

Hole

PhotonsEg

Applied electric field

Intrinic mechanism

Eex

More energy levels in the bandgap are done by doping at low concentrations, typically < 10-8 like AS doped SI. This is called extrinsic. For long wavelength detectors like Si:As.

Bandgaps for various detector materials:

Photovoltaic effect



x Eg (eV) c (m)

0.196 .09 14

0.21 .12 10

0.295 .25 5

0.395 .41 3

0.55 .73 1.7

0.7 1.0 1.24

Tunable Bandgap - A great property of Mer-Cad-Tel

Hg1-xCdxTe

Modify ratio of Mercury and Cadmium

to “tune” the bandgap energy



Tunable Bandgap - A great property of Mer-Cad-Tel



Detectors used in Astronomy are made out of semiconductors

The Photoelectric effect is the basic principle

To avoid thermal excitation detectors need to be cooled

The photons can generate photo-electrons in conduction band of semiconductors

The material of semiconductors determines band gap energy which determines the cutoff wavelength of the detector material

The photo-electrons needed to be transferred, be amplified, and eventually bedigitized.

Semiconductor summary



Light sensitive materialis electrically partitionedinto a 2-D array of pixels

(each pixel is a 3-D volume)Photons > Electrons

Solid state electronicsthat amplify and read

out the chargex

y

z

• Intensity image is generated by collecting photo charge generated in 3-D volume into 2-D array of pixels.

• Optical and IR focal plane arrays both collect charges via electric fields.

• In the z-direction, optical and IR use a p-n junction to “sweep” charge toward pixel collection nodes.

Detector architecture (CCD and CMOS)

Photons



Absorption depth of SI and HgCdTe

Silicon (indirect bandgap)HgCdTe (direct bandgap)

Absorption depth = The depth of detector material that absorbs 63.2% of the radiation

1 absorption depth(s) 63.2% of light absorbed2 86.5%3 95.0% 4 98.2%

For high QE, thickness of detector material should be ≥ 3 absorption depths

Indirect bandgap material: Electron needs change in momentum in addition to an energy change !

IR detector material is very thin 10 to 15 micron, SI detector can be very thick (i.e. 300 microns)



CCD needs charge transfer towards amplifier

Red electrodes high potential and green the lowpotentials.

A pixel is the region between two channel stops

During the exposure two gates are held at high potentials to integrate charge in the pixel

Pixels are read after the integration

CCD pixel share the same amplifier

3 phase CCD Hybrid CMOS/IR

Detector material hybridized to SI multiplexer (optical or IR material)

No „charge coupling‟

Indium interconnects are used

Charge to voltage conversion takes place in parallel at the sense node of each pixel

CMOS have amplifier per unit cell

Pixels can be read during the integration



Then the voltage gets amplified by a MOSFET transistor.

Frontside (front-illuminated ) CCD

Poly gates (3 phase structure CCD)

p- - epitaxial layer

Thinned (back-illuminated ) CCD

p+- substrate

p- - epitaxial layer

Photosensitive volume (20µm)

Photons

Photons

n - buried channel

Principle of CCD Sensors

BI CCDs have the best spectral response available

The CCD is inverted, the bulk silicon ground down and Anti-Reflection (AR) coating is added. A number

of optimized AR coating options are available

During the integration time charge is collected underone or two of the gates.

After an exposure the charge needs to be moved towards the output structure of the CCD. A simple scheme of clock pulses is applied to the gates to move the charge from one pixel to the next. Such clock cycles are repeated to readout an entire N-pixellinear registers for parallel or serial movement



RESET

ROW

SELECT

COLUMN

SELECT

UNIT CELLSOURCE

FOLLOWER

OFF CHIP

LOAD

RESISTOR

INDIUMBUMP

NARROW

BANDGAP

DIODE ARRAY

REVERSE BIAS

VOLTAGE ~ +500 mV

Si CMOS

MULTIPLEXER

Principle of Hybrid Active Pixel Sensors

Structure

Silicon readout multiplexer

Narrow band-gap infrared diode array

Hybridization with In bumps

Operation

charge diode capacity by reverse bias voltage

floating capacity is discharged by absorbed photons

Read voltage across diode capacity several times during integration by addressing unit cell source follower



Difference between optical and IR Hybrid Active Pixel Sensors

SI-MULTIPLEXER (ROIC)

Vsub (bias voltage)

hv-Photons

fully depleted bulk (SI)

Implant

oxide

AR-Coating on surface

E-Field

Indium bump

metal grid

Typical SI-PIN ARRAY

Typical SI Hybrid with fully depleted bulk

SI-MULTIPLEXER (ROIC)

hv-Photons

bulk (IR)

Region of depletion

AR-Coating on surface

E-Field=0Indium bump

Vdet contact

Typical IR - ARRAY

Typical IR array with per pixel depleted bulk

Implant boron ions to form n-on-p junctions

SI-PIN array is a fully depleted bulk detector

IR array is a per pixel depleted detector.



Summary of detector architecture

CCD CMOS

Pixel

Charge generation &charge integration

Charge generation, charge integration & charge-to-voltage

conversion

+

PhotodiodePhotodiode Amplifier

Array Readout

Charge transfer from pixel to pixel

Multiplexing of pixel voltages: Successively connect amplifiers to

common bus

Sensor Output

Output amplifier performs charge-to-voltage conversion

Various options possible:

- no further circuitry (analog out)- add. amplifiers (analog output)- A/D conversion (digital output)



No

ise

(RM

S)

in

[A

DU

]

Signal level in [ADU]

Note: Both axis are in logaritmic scale

PHOTON TRANSFER CURVE NOISE

Read noise Shot noise Fixed pattern noise

well capacity

slope 0

slope 0.5

slope 1

How do I know how many electrons the pixel has collected since we only record digital values ? => The photon transfer curve

Measurement of detector parameters such as noise, system gain, full well capacity, quantum efficiency, dark current, sensitivity and linearity are usually covered using the photon transfer curve.

A photon transfer curve has three different noise regimes:

1. readnoise2. shot noise 3. fixed pattern noise.

Photon Transfer (1)



Read noise is the noise associated with the detectors output amplifier and the readout electronics (i.e. its signal processing, digitization etc.). This is the intrinsic system noise of a dark frame or image (no light). It is independent of the photons or input signal. The slope is 0 on a logarithmic scale.

Shot noise occurs when the input signal increases and the noise of the detector is dominated by shot noise. Shot noise is proportional to the square root of that signal. The slope is 0.5 on a logarithmic scale.

Fixed Pattern noise arises at high levels of illumination. This noise results from differences in sensitivity of pixels. This is also called the Pixel Response Nonuniformity (PRNU).

Due to processing and mask alignment variations during manufacture each pixel has a slightly different charge collection capacity and responsivity. This noise is proportional to the number of photons. The slope is 1 on a logarithmic scale.

Photon Transfer (2)



100 200 300 400 500 600 700 80050

100

150

200

250

Signal [ADU]

variance o

f th

e s

ignal [A

DU

]2

Photon transfer curve HyViSI detector

data hyvisi

linear fit

gain = 3.55 +/- 0.034 e/ADU

Photon Transfer (3) Example:The photon transfer curve plots read noise as a function of the signal for an area of n by n pixels in a frame.

Np

SS

S

pNi

i

darki

1

To obtain the y-axis of the curve, the variance is computed. The variance is the square of the standard deviation of a single observation from the mean of the pixels.

Np

SS

pNi

i

i

1

2

2

To obtain the x-axis of the curve one computes the mean, dark subtracted signal S. That is

where Si is the signal value of the ith pixel and

Np is the number of pixels in the n by n pixel area.

Sdark is the signal of a dark frame taken from the same data set.



FE-55FE-55 is a radioactive source that emits X rays at three energy levels (5.9 KeV (Mn K line), weaker peak at 6.5 KeV (Mn K) and the third at 4.12KeV (K escape line).

HyViSI FE-55 histogram: Conversion factor 1.65 e/ADU

FE-55 events on the detector (120s

integration time)

FE-55 source installed on the window

When these Xrays are absorbed by silicon they produce large photoelectron events K 1620 electrons, K 1778 electrons and the K escape peak 1133 electrons.

The K line was used to calibrate the conversion gain



CMOS Detectors readout scheme

ReadReadReset

Time

Pixel 1 Pixel 1000

100 ms 100 ms T int 100 ms

To define the exposure time IR detectors do not require a shutter. If shutters are used those would have to be cold and operate very fast due to short exposure times in the infrared due to high background radiation.

IR detectors are read out non-destructive ( sampling does not alter the charge on the photodiode junction). When a detector is reset the signal shifts to the pedestal level. Then the diode discharges either by photocurrent or dark current. Resets are done usually pixel by pixel.

The following sample/reset modes are mainly used in

astronomy (Diagram on a pixel by pixel basis)



IR Detectors readout scheme

Time [s]

Vo

ltag

e ac

ross

the

dio

de

[V]

Reset Discharge due to photons

Reset

Read

Single or Uncorrelated Sampling

Sign

al l

evel

Time [s]

Vo

ltag

e ac

ross

the

dio

de

[V]


Reset

Read

Correlated Double Sampling

Sign

al l

evel

Read

Single (reset read) or uncorrelated Sampling

Cannot remove KTC noise or drifts in the detector but can

measure saturation or full well capacity of the detector

pixels (use also for dark current measurements by not

resetting the device). Provides high dynamic range.

KTC noise = drifts in voltage due to Temp effects.

This mode removes KTC noise but cannot detect saturation of

the pixels. It is the standard readout mode.

Correlated Double Sampling (CDS)




Time [s]

Vo

ltag

e ac

ross

the

dio

de

[V]


Reset

Read

Fowler (reset read read) Sampling

Sign

al l

evel

ReadRead

Read

Time [s]

Vo

ltag

e ac

ross

the

dio

de

[V]


Reset

Read

Up the ramp Sampling

Sig

nal

leve

l

Read

Read

Read

Read

Read

Read

Readnoise decreases as n

1

with n being the numbers of samples.

Is better in readnoise limited conditions than DC.

Saturation not known.

Up-the-Ramp Sampling

Fit line to get the mean flux rate = slope.

This mode is good if some pixels saturate before the end

of the exposure time.

Fowler (reset-read-read) Sampling




Non-destructive readout enables reduction of noise from multiple samples

Theory

Measured

H2RG array

2.5 micron cutoff

Temperature = 77K



OPTICAL DETECTORS for imaging and spectroscopy

i.e. ESO‟s Scientific CCDs

2k x 4k, 15 micron pixels, 3-side (and 4-side) buttable , now also 4kx4k CCDsDark current < 1 electron/pixel/hrHigh speed, low noise amplifiers (2 e- at 50 kps, 5 e- at 625 kps)Readout speed up to 1 Million pixels per secondTypical Charge transfer efficiency: 0.999999 (six 9‟s)Very flat (less than 20 micron peak-to-valley)Excellent cosmetic quality ( 4 bad columns)Full well capacity: 130 000 to 225000 electrons

Many manufacturers produce these devices, with different spectral response:

e2v, MIT Lincoln Laboratory, etc.



Typical quantum efficiency of thinned E2V and MIT/LL CCDs

Quantum efficiency

0

10

20

30

40

50

60

70

80

90

100

300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050 1100

Wavelength [nm]

QE

[%]

EEV CCD-44 [STING] MIT/LL CCID-20 [NIGEL] MIT/LL Thick



Examples of CCD detectors systems

…from single detectors to very big mosaics…..

Single E2V 2kx4k CCD

Wide Field Imager 8k x 8k mosaic, 72 million pixels

Mosaic of two E2V 2kx4k CCD

E2V 4kx4k CCD231 with graded AR Coating



and even bigger…..

OmegaCAM detector mosaic

32 CCDs - 16 x 16 k - 1x1° FOV + 4 tracker - 288 million pixels

(courtesy: Olaf Iwert ,ESO)



SI-PIN/Visible hybrid Hawaii2RG detector

A silicon pin hybrid detector has close synergy with IR (HgCdTe) detectors.

It is a complementary metal oxide semiconductor (CMOS) alternative to charge coupled devices (CCDs) for photons at optical wavelength.

2k x 2k format with 18 micron pixels



HyViSI quantum efficiency (compared to CCDs)

The HyViSI detector outperforms all CCDs above 500 nm and shows a higher overall QE compared to the CCDs.

The e2v astro is a curve providedby e2v for a broad band deepdepletion device.

The green curve the QE for a 2layer AR coating of the deepdepletion CCD.

The blue curve is the QE of theCCD currently installed inGiraffe at the VLT.

Red curve is a IR Hawaii2RGHgCdTe detector

Comparision of QE: CCDs - HyViSI - HgCdTe Hawaii2RG

0

10

20

30

40

50

60

70

80

90

100

300 400 500 600 700 800 900 1000 1100 1200 1300

Wavelength [nm]

QE

[%

]

Existing e2v 44-82 Bruce e2v astro BB DD

HyViSI - 180K e2v 2 layer AR coating (b) DD

HgCdTe Hawaii2RG with 2.5 micron cutoff



Infrared detectors used in Astronomy

Infrared astronomy is currently benefiting from three different technologiesproviding high performance hybrid active pixel sensors:

In the near infrared from 1 to 5 m two technologies:

InSb and Hg(1-x)CdxTe grown by LPE or MBE on Al2O3, Si or CdZnTe substrates.

The width of the band-gap of the alloy Hg(1-x)CdxTe can be tuned by varying the composition x of the alloy. In this way the cut-off wavelength c of the sensor can be changed as explained before.

In the mid infrared spectral range from 8 to 28 m:

Blocked impurity band Si:As arrays



Present IR arrays

Rockwell 1024x1024 2.5m HgCdTe detector array4 Quadrant architecture4 Output amplifiers18.5 m pixelsLPE HgCdTe on sapphire (PACE-1)Use of external JFETs possible

Hawaii 1Quantum efficiency (70% - 80%)Dark current 0.01 e-/s (65K)Read noise about 10 - 15 e- rms CDSResidual image effectSome multiplexer glowFringing500 ms – 1 s frame time

1024x1024 InSb detector array4 Quadrant architecture32 Output amplifiers27 m pixelsThinned, AR coated InSbThree generations of multiplexersFrame time ~ 70 ms

Quantum efficiency high (70% - 90%)Dark current 0.01 - 1.0 e-/sRead noise about 40 e- rms CDS,

10 e- rms Fowler samplingCharge capacity 200,000 e-Residual image effect



Present IR arrays

2048 x 2048 resolution with 18 µm square pixelsClose buttable package 1, 4, or 32 output mode selectableSlow mode (100 kHz) and fast mode (5 MHz with additional column buffers) selectable, both usable with internal and external buffers

Most Sophisticated ROIC Yet Developed for Astronomy

Noise: 8 to 15 electrons for a normal DC readQE (array mean ) > 80 %Dark current < 0.006 e/s/pixel at 77K and 2.5 um cutoffSpectral range 0.3 - 5.3 um Guide mode and reference pixels

Number ofoutputs

Frame time in slowmode

Frame time infast mode

1 42 s 840 ms

4 10.5 s 210 ms

32 1.3 s 26 ms

Hawaii 2RG



STScI

Fowler sampling:number of readouts nproportional to integrationtime: 825 ms/readout

for 256 Fowler pairs 3 e- rms on IR pixels 1.8 e- rms on reference pixels

shielding multiplexer glowvery efficient

large number of nondestructive readouts possible with 32 channels

Hawaii 2RG noise performance

(courtesy: Gert Finger,ESO)



Hawaii 2RG QE performance

(courtesy: Gert Finger,ESO)



The CRIRES 1024 x 4096 pixels Aladdin InSb focal plane array

Four Aladdin 1Kx1K InSb arrays

Cryogenic Echelle Spectrographcurvature AO: 0.1 arcsec / pixel512 pixels in spatial directionHigh resolution R=100000 echelleprism predisperser for order sorting and photon background suppression

Examples of IR detectors systems



The CRIRES 1024 x 4096 pixels Aladdin InSb focal plane array

A new 3 side quasi buttable package for the Aladdin II /III

(ESO development)

Buttable package

AlN chip carrier

Aladdin III in new package



HAWK-I – HgCdTe Array Wide field K-band Imager for the VLT

Wavelength range: 0.85 - 2.5 µm

Mosaic out of 4 Hawaii 2RG MBE detectors, 128 parallel channel system



VIRGO 16 x 2Kx2K HgCdTe mosaic for VISTA (4m survey Telescope)

• VISTA built by RAL & UKATC• FOV 1.65 degrees• 2Kx2K HgCdTe grown by LPE

on CdZnTe substrate (VIRGO)• Pixel size 20 m• 16 parallel outputs• Pixel rate 400 KHz• Frame rate 1.5 Hz• 3-side buttable• Reference cells included in video

data stream



The ESO baseline controller for CCDs and IR detectors (NGC)

• NGC is a modular system for IR detector and CCD readout with a Back-end, a basic Front-end unit containing a complete four channel system on one card and additional boards like 32 channel ADC units and more...

• There is no processor, no parallel inter-module data bus on the front-end side. Advanced FPGA link technology is used to replace conventional logic.

• Connection between Back and Front-end with high speed fiber links at 2.5GBit/s

• Connection between Front-end modules with high speed copper links at 2.5GBit/s.

FrontEnd

BackEnd



SIDECAR ASIC

SIDECAR™- system image, digitizing, enhancing, controlling, and retrieving -

ASIC - Application Specific Integrated Circuit -

The ASIC is a controller on a single Chip designed for use in all Teledyne Imaging Sensors (former Rockwell) FPAs including 2048 x 2048 HAWAII-2RG™



SIDECAR ASIC



ASIC @ ESO – LCC package



JADE card on the outside

ESO- ASIC cryogenic setup inside cryostat


Vincent van Gogh - Starlight Over The Rhone

Basic principles of photon detectors used in Astronomy...May 13, 2009 · Reinhold Dorn -Basic...

Documents

Transcript of Basic principles of photon detectors used in Astronomy...May 13, 2009 · Reinhold Dorn -Basic...